Update app.py

app.py

@@ -5,76 +5,87 @@ import requests
 import gradio as gr
 from huggingface_hub import hf_hub_download
 
-MODEL_REPO = "bartowski/Phi-3-mini-4k-instruct-GGUF"
-MODEL_FILE = "Phi-3-mini-4k-instruct-Q4_K_M.gguf"
+MODEL_REPO = "bartowski/Qwen_Qwen3.5-9B-GGUF"
+MODEL_FILE = "Qwen3.5-9B-Q4_K_M.gguf"
 
 LLAMA_DIR = "llama.cpp"
 SERVER_PORT = "8000"
 
-def setup_model():
-    path = hf_hub_download(
+
+def download_model():
+    return hf_hub_download(
         repo_id=MODEL_REPO,
         filename=MODEL_FILE
     )
-    return path
+
 
 def setup_llama():
     if not os.path.exists(LLAMA_DIR):
         subprocess.run(
-            ["git","clone","https://github.com/ggerganov/llama.cpp"],
+            ["git", "clone", "https://github.com/ggerganov/llama.cpp"],
             check=True
         )
 
     subprocess.run(
-        ["make","-j"],
+        ["make", "-j"],
         cwd=LLAMA_DIR,
         check=True
     )
 
+
 def start_server(model_path):
-    subprocess.Popen([
-        "./server",
-        "-m", model_path,
-        "--port", SERVER_PORT,
-        "--ctx-size", "4096",
-        "--api"
-    ], cwd=LLAMA_DIR)
+    subprocess.Popen(
+        [
+            "./server",
+            "-m", model_path,
+            "--port", SERVER_PORT,
+            "--ctx-size", "4096",
+            "--api"
+        ],
+        cwd=LLAMA_DIR
+    )
+
 
 def wait_for_server():
     url = f"http://localhost:{SERVER_PORT}/health"
-    for _ in range(30):
+    for _ in range(60):
         try:
             requests.get(url, timeout=1)
             return
         except:
             time.sleep(2)
 
+
 def chat(prompt):
     r = requests.post(
         f"http://localhost:{SERVER_PORT}/v1/chat/completions",
         json={
-            "model": "local",
+            "model": "qwen3.5-9b",
             "messages": [
-                {"role":"user","content": prompt}
+                {"role": "user", "content": prompt}
             ]
         },
-        timeout=120
+        timeout=300
     )
 
     data = r.json()
     return data["choices"][0]["message"]["content"]
 
-model_path = setup_model()
+
+model_path = download_model()
+
 setup_llama()
+
 start_server(model_path)
+
 wait_for_server()
 
+
 ui = gr.Interface(
     fn=chat,
-    inputs=gr.Textbox(lines=4, placeholder="Ask something..."),
+    inputs=gr.Textbox(lines=5, placeholder="Ask Operon something..."),
     outputs="text",
-    title="Operon Dev LLM",
-    description="Local llama.cpp server running on Hugging Face Space"
+    title="Operon Dev LLM (Qwen3.5-9B)",
 )
 
 ui.launch(server_name="0.0.0.0", server_port=7860)